The association of particular patterns of dysfunctional familial attachments with HIV-infection-risk through drug use and sexual activities will be examined based on data gathered over a two year period from approximately 1 000 youths adjudicated to the Virginia Learning Centers system. The underlying theoretical hypothesis informing this study is that the HIV-risk behaviors of drug use and unsafe sex are associated with particular styles or patterns of dysfunctional familial attachments and bonding characteristics. We are also interested in investigating the two other forms of risk-taking behaviors in adolescence -- suicidal ideation and behaviors, and aggressive behaviors -- to enable us to construct a model of attachment and behavioral characteristics associated with the HIV-risk behaviors of drug use and unsafe sex acts. Study data will include standardized and validated questionnaires to assess attachment and bonding characteristics of the youths with their families; psychosocial characteristics that have been identified as relevant to risk-taking; and four types of risk-taking behaviors: drug use, unsafe sexual activities, suicidal ideation/behaviors and aggressive behaviors. The study will also assess the adolescents' knowledge of and attitudes about risk of HIV-infection and the consequences of infection. A stratified random sample of 1 60 youths will be drawn from the complete study sample to test six specific main effect hypotheses of association of (1) risk of HIV infection due to drug use and/or sexual activities, and (2) familial attachment characteristics, in particular; and (3) other psychosocial and behavioral characteristics in general. The sample will be stratified by three variables: drug use, unsafe sexual activities and sex. This stratification will yield four risk groups with 25 males and 1 5 females in each group. Multivariate analytic techniques will be used to derive a structured mathematical model predicting membership in the four defined groups, and testing the extent to which the attachment variables mediate the relationship of other variables with drug use and sexual risk-taking. The effect of age and race on the observed relationships will also be controlled, using multivariate techniques. The utility of the model will be confirmed by using a discriminant functions analysis based on the derived model to classify those subjects who were not part of the random sample into the risk categories. As a result of this study, those involved with the treatment and rehabilitation of incarcerated adolescents should have more information on the probability of co-occurrence of the various types of risktaking behaviors and on the attachment and bonding characteristics associated with these behaviors. This information can help determine specific priorities and strategies for primary, secondary and tertiary prevention in these settings. In addition, by modeling the psychosocial and familial factors distinguishing various patterns of risk-taking behaviors, this study can contribute to understanding the etiology of these problems and defining positive outcomes in more substantive ways than simple behavioral modification and symptom reduction.